GUOTAI HAITONG SECURITIES·2026-01-27 10:59Quantitative Models and Construction Methods 1. Model Name: Large Order Transaction Amount Ratio - Model Construction Idea: This model tracks the buying behavior of large funds by calculating the proportion of large order transaction amounts to the total daily transaction amount[7] - Model Construction Process: 1. Use tick-by-tick transaction data to identify buy and sell orders based on the sequence numbers of bids and asks 2. Filter transactions by order size to identify large orders 3. Calculate the proportion of large buy order transaction amounts to the total daily transaction amount - Formula: $ \text{Large Order Transaction Amount Ratio} = \frac{\text{Large Buy Order Transaction Amount}}{\text{Total Daily Transaction Amount}} $ - Model Evaluation: This model effectively captures the buying behavior of large funds, providing insights into market dynamics[7] 2. Model Name: Net Active Buy Amount Ratio - Model Construction Idea: This model measures investors' active buying behavior by calculating the net active buy amount as a proportion of the total daily transaction amount[7] - Model Construction Process: 1. Use tick-by-tick transaction data to classify each transaction as either active buy or active sell based on the buy/sell flag 2. Calculate the net active buy amount by subtracting the active sell amount from the active buy amount 3. Compute the ratio of the net active buy amount to the total daily transaction amount - Formula: $ \text{Net Active Buy Amount Ratio} = \frac{\text{Active Buy Amount} - \text{Active Sell Amount}}{\text{Total Daily Transaction Amount}} $ - Model Evaluation: This model provides a clear view of active investor sentiment and their willingness to buy[7] --- Model Backtesting Results 1. Large Order Transaction Amount Ratio - Top 5 Stocks by 5-Day Average: - Jianghua Microelectronics: 96.1% (98.8% time-series percentile)[9] - Fenglong Co.: 94.9% (92.1% time-series percentile)[9] - Ningbo Port: 86.4% (94.2% time-series percentile)[9] - Hongta Securities: 86.4% (100.0% time-series percentile)[9] - Chongqing Steel: 86.1% (77.0% time-series percentile)[9] 2. Net Active Buy Amount Ratio - Top 5 Stocks by 5-Day Average: - Liaogang Co.: 25.5% (98.8% time-series percentile)[10] - Rong'an Real Estate: 22.8% (99.6% time-series percentile)[10] - Beichen Industrial: 21.2% (100.0% time-series percentile)[10] - Angang Steel: 20.6% (100.0% time-series percentile)[10] - Anyang Steel: 20.3% (100.0% time-series percentile)[10] 3. Broad-Based Indices - Large Order Transaction Amount Ratio (5-Day Average): - Shanghai Composite Index: 73.2% (47.7% percentile)[12] - CSI 300: 72.2% (28.0% percentile)[12] - CSI 500: 74.0% (97.9% percentile)[12] - Net Active Buy Amount Ratio (5-Day Average): - Shanghai Composite Index: -0.6% (56.4% percentile)[12] - CSI 300: -7.4% (70.8% percentile)[12] - CSI 500: 4.9% (95.1% percentile)[12] 4. Industry-Level Results - Top Industries by Large Order Transaction Amount Ratio (5-Day Average): - Steel: 78.6% (45.7% percentile)[13] - Coal: 77.7% (49.8% percentile)[13] - Media: 77.6% (70.4% percentile)[13] - Top Industries by Net Active Buy Amount Ratio (5-Day Average): - Steel: 7.1% (19.3% percentile)[13] - Nonferrous Metals: 5.0% (9.1% percentile)[13] - Media: 3.2% (45.7% percentile)[13] 5. ETFs - Top 5 ETFs by Large Order Transaction Amount Ratio (5-Day Average): - Huatai-PineBridge CSI A500 ETF: 93.0% (96.7% percentile)[15] - Harvest CSI 300 ETF: 92.3% (100.0% percentile)[15] - E Fund CSI 300 ETF: 92.1% (99.2% percentile)[15] - Top 5 ETFs by Net Active Buy Amount Ratio (5-Day Average): - Penghua CSI Sub-Sector Chemical Industry ETF: 22.7% (92.2% percentile)[16] - Bosera CSI Sub-Sector Chemical Industry ETF: 18.2% (99.2% percentile)[16] - Huaxia GEM Artificial Intelligence ETF: 16.9% (96.6% percentile)[16]